4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
13 #include "parse-options.h"
16 #ifndef NO_EXTERNAL_GREP
18 #define NO_EXTERNAL_GREP 0
20 #define NO_EXTERNAL_GREP 1
24 static char const * const grep_usage
[] = {
25 "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
29 static int grep_config(const char *var
, const char *value
, void *cb
)
31 struct grep_opt
*opt
= cb
;
33 if (!strcmp(var
, "color.grep")) {
34 opt
->color
= git_config_colorbool(var
, value
, -1);
37 if (!strcmp(var
, "color.grep.external"))
38 return git_config_string(&(opt
->color_external
), var
, value
);
39 if (!strcmp(var
, "color.grep.match")) {
41 return config_error_nonbool(var
);
42 color_parse(value
, var
, opt
->color_match
);
45 return git_color_default_config(var
, value
, cb
);
49 * git grep pathspecs are somewhat different from diff-tree pathspecs;
50 * pathname wildcards are allowed.
52 static int pathspec_matches(const char **paths
, const char *name
)
55 if (!paths
|| !*paths
)
57 namelen
= strlen(name
);
58 for (i
= 0; paths
[i
]; i
++) {
59 const char *match
= paths
[i
];
60 int matchlen
= strlen(match
);
61 const char *cp
, *meta
;
64 ((matchlen
<= namelen
) &&
65 !strncmp(name
, match
, matchlen
) &&
66 (match
[matchlen
-1] == '/' ||
67 name
[matchlen
] == '\0' || name
[matchlen
] == '/')))
69 if (!fnmatch(match
, name
, 0))
71 if (name
[namelen
-1] != '/')
74 /* We are being asked if the directory ("name") is worth
77 * Find the longest leading directory name that does
78 * not have metacharacter in the pathspec; the name
79 * we are looking at must overlap with that directory.
81 for (cp
= match
, meta
= NULL
; cp
- match
< matchlen
; cp
++) {
83 if (ch
== '*' || ch
== '[' || ch
== '?') {
89 meta
= cp
; /* fully literal */
91 if (namelen
<= meta
- match
) {
92 /* Looking at "Documentation/" and
93 * the pattern says "Documentation/howto/", or
94 * "Documentation/diff*.txt". The name we
95 * have should match prefix.
97 if (!memcmp(match
, name
, namelen
))
102 if (meta
- match
< namelen
) {
103 /* Looking at "Documentation/howto/" and
104 * the pattern says "Documentation/h*";
105 * match up to "Do.../h"; this avoids descending
106 * into "Documentation/technical/".
108 if (!memcmp(match
, name
, meta
- match
))
116 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
, int tree_name_len
)
120 enum object_type type
;
121 char *to_free
= NULL
;
124 data
= read_sha1_file(sha1
, &type
, &size
);
126 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
129 if (opt
->relative
&& opt
->prefix_length
) {
130 static char name_buf
[PATH_MAX
];
132 int name_len
= strlen(name
) - opt
->prefix_length
+ 1;
135 name
+= opt
->prefix_length
;
137 if (ARRAY_SIZE(name_buf
) <= name_len
)
138 cp
= to_free
= xmalloc(name_len
);
141 memcpy(cp
, name
, tree_name_len
);
142 strcpy(cp
+ tree_name_len
,
143 name
+ tree_name_len
+ opt
->prefix_length
);
147 hit
= grep_buffer(opt
, name
, data
, size
);
153 static int grep_file(struct grep_opt
*opt
, const char *filename
)
160 if (lstat(filename
, &st
) < 0) {
163 error("'%s': %s", filename
, strerror(errno
));
167 return 0; /* empty file -- no grep hit */
168 if (!S_ISREG(st
.st_mode
))
170 sz
= xsize_t(st
.st_size
);
171 i
= open(filename
, O_RDONLY
);
174 data
= xmalloc(sz
+ 1);
175 if (st
.st_size
!= read_in_full(i
, data
, sz
)) {
176 error("'%s': short read %s", filename
, strerror(errno
));
182 if (opt
->relative
&& opt
->prefix_length
)
183 filename
+= opt
->prefix_length
;
184 i
= grep_buffer(opt
, filename
, data
, sz
);
189 #if !NO_EXTERNAL_GREP
190 static int exec_grep(int argc
, const char **argv
)
200 execvp("grep", (char **) argv
);
203 while (waitpid(pid
, &status
, 0) < 0) {
208 if (WIFEXITED(status
)) {
209 if (!WEXITSTATUS(status
))
218 #define push_arg(a) do { \
219 if (nr < MAXARGS) argv[nr++] = (a); \
220 else die("maximum number of args exceeded"); \
224 * If you send a singleton filename to grep, it does not give
225 * the name of the file. GNU grep has "-H" but we would want
226 * that behaviour in a portable way.
228 * So we keep two pathnames in argv buffer unsent to grep in
229 * the main loop if we need to do more than one grep.
231 static int flush_grep(struct grep_opt
*opt
,
232 int argc
, int arg0
, const char **argv
, int *kept
)
235 int count
= argc
- arg0
;
236 const char *kept_0
= NULL
;
240 * Because we keep at least 2 paths in the call from
241 * the main loop (i.e. kept != NULL), and MAXARGS is
242 * far greater than 2, this usually is a call to
243 * conclude the grep. However, the user could attempt
244 * to overflow the argv buffer by giving too many
245 * options to leave very small number of real
246 * arguments even for the call in the main loop.
249 die("insanely many options to grep");
252 * If we have two or more paths, we do not have to do
253 * anything special, but we need to push /dev/null to
254 * get "-H" behaviour of GNU grep portably but when we
255 * are not doing "-l" nor "-L" nor "-c".
259 !opt
->unmatch_name_only
&&
261 argv
[argc
++] = "/dev/null";
268 * Called because we found many paths and haven't finished
269 * iterating over the cache yet. We keep two paths
270 * for the concluding call. argv[argc-2] and argv[argc-1]
271 * has the last two paths, so save the first one away,
272 * replace it with NULL while sending the list to grep,
273 * and recover them after we are done.
276 kept_0
= argv
[argc
-2];
281 status
= exec_grep(argc
, argv
);
285 * Then recover them. Now the last arg is beyond the
286 * terminating NULL which is at argc, and the second
287 * from the last is what we saved away in kept_0
289 argv
[arg0
++] = kept_0
;
290 argv
[arg0
] = argv
[argc
+1];
295 static void grep_add_color(struct strbuf
*sb
, const char *escape_seq
)
297 size_t orig_len
= sb
->len
;
299 while (*escape_seq
) {
300 if (*escape_seq
== 'm')
301 strbuf_addch(sb
, ';');
302 else if (*escape_seq
!= '\033' && *escape_seq
!= '[')
303 strbuf_addch(sb
, *escape_seq
);
306 if (sb
->len
> orig_len
&& sb
->buf
[sb
->len
- 1] == ';')
307 strbuf_setlen(sb
, sb
->len
- 1);
310 static int external_grep(struct grep_opt
*opt
, const char **paths
, int cached
)
312 int i
, nr
, argc
, hit
, len
, status
;
313 const char *argv
[MAXARGS
+1];
314 char randarg
[ARGBUF
];
315 char *argptr
= randarg
;
318 if (opt
->extended
|| (opt
->relative
&& opt
->prefix_length
))
328 if (opt
->regflags
& REG_EXTENDED
)
330 if (opt
->regflags
& REG_ICASE
)
332 if (opt
->binary
== GREP_BINARY_NOMATCH
)
334 if (opt
->word_regexp
)
338 if (opt
->unmatch_name_only
)
340 if (opt
->null_following_name
)
341 /* in GNU grep git's "-z" translates to "-Z" */
345 if (opt
->post_context
|| opt
->pre_context
) {
346 if (opt
->post_context
!= opt
->pre_context
) {
347 if (opt
->pre_context
) {
349 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
350 "%u", opt
->pre_context
) + 1;
351 if (sizeof(randarg
) <= len
)
352 die("maximum length of args exceeded");
356 if (opt
->post_context
) {
358 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
359 "%u", opt
->post_context
) + 1;
360 if (sizeof(randarg
) <= len
)
361 die("maximum length of args exceeded");
368 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
369 "%u", opt
->post_context
) + 1;
370 if (sizeof(randarg
) <= len
)
371 die("maximum length of args exceeded");
376 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
378 push_arg(p
->pattern
);
381 struct strbuf sb
= STRBUF_INIT
;
383 grep_add_color(&sb
, opt
->color_match
);
384 setenv("GREP_COLOR", sb
.buf
, 1);
387 strbuf_addstr(&sb
, "mt=");
388 grep_add_color(&sb
, opt
->color_match
);
389 strbuf_addstr(&sb
, ":sl=:cx=:fn=:ln=:bn=:se=");
390 setenv("GREP_COLORS", sb
.buf
, 1);
394 if (opt
->color_external
&& strlen(opt
->color_external
) > 0)
395 push_arg(opt
->color_external
);
400 for (i
= 0; i
< active_nr
; i
++) {
401 struct cache_entry
*ce
= active_cache
[i
];
404 if (!S_ISREG(ce
->ce_mode
))
406 if (!pathspec_matches(paths
, ce
->name
))
409 if (name
[0] == '-') {
410 int len
= ce_namelen(ce
);
411 name
= xmalloc(len
+ 3);
412 memcpy(name
, "./", 2);
413 memcpy(name
+ 2, ce
->name
, len
+ 1);
416 if (MAXARGS
<= argc
) {
417 status
= flush_grep(opt
, argc
, nr
, argv
, &kept
);
425 } while (i
< active_nr
&&
426 !strcmp(ce
->name
, active_cache
[i
]->name
));
427 i
--; /* compensate for loop control */
431 status
= flush_grep(opt
, argc
, nr
, argv
, NULL
);
439 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
,
440 int external_grep_allowed
)
446 #if !NO_EXTERNAL_GREP
448 * Use the external "grep" command for the case where
449 * we grep through the checked-out files. It tends to
450 * be a lot more optimized
452 if (!cached
&& external_grep_allowed
) {
453 hit
= external_grep(opt
, paths
, cached
);
459 for (nr
= 0; nr
< active_nr
; nr
++) {
460 struct cache_entry
*ce
= active_cache
[nr
];
461 if (!S_ISREG(ce
->ce_mode
))
463 if (!pathspec_matches(paths
, ce
->name
))
466 * If CE_VALID is on, we assume worktree file and its cache entry
467 * are identical, even if worktree file has been modified, so use
468 * cache version instead
470 if (cached
|| (ce
->ce_flags
& CE_VALID
)) {
473 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
, 0);
476 hit
|= grep_file(opt
, ce
->name
);
480 } while (nr
< active_nr
&&
481 !strcmp(ce
->name
, active_cache
[nr
]->name
));
482 nr
--; /* compensate for loop control */
485 free_grep_patterns(opt
);
489 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
490 struct tree_desc
*tree
,
491 const char *tree_name
, const char *base
)
495 struct name_entry entry
;
497 int tn_len
= strlen(tree_name
);
498 struct strbuf pathbuf
;
500 strbuf_init(&pathbuf
, PATH_MAX
+ tn_len
);
503 strbuf_add(&pathbuf
, tree_name
, tn_len
);
504 strbuf_addch(&pathbuf
, ':');
505 tn_len
= pathbuf
.len
;
507 strbuf_addstr(&pathbuf
, base
);
510 while (tree_entry(tree
, &entry
)) {
511 int te_len
= tree_entry_len(entry
.path
, entry
.sha1
);
513 strbuf_add(&pathbuf
, entry
.path
, te_len
);
515 if (S_ISDIR(entry
.mode
))
516 /* Match "abc/" against pathspec to
517 * decide if we want to descend into "abc"
520 strbuf_addch(&pathbuf
, '/');
522 down
= pathbuf
.buf
+ tn_len
;
523 if (!pathspec_matches(paths
, down
))
525 else if (S_ISREG(entry
.mode
))
526 hit
|= grep_sha1(opt
, entry
.sha1
, pathbuf
.buf
, tn_len
);
527 else if (S_ISDIR(entry
.mode
)) {
528 enum object_type type
;
529 struct tree_desc sub
;
533 data
= read_sha1_file(entry
.sha1
, &type
, &size
);
535 die("unable to read tree (%s)",
536 sha1_to_hex(entry
.sha1
));
537 init_tree_desc(&sub
, data
, size
);
538 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
542 strbuf_release(&pathbuf
);
546 static int grep_object(struct grep_opt
*opt
, const char **paths
,
547 struct object
*obj
, const char *name
)
549 if (obj
->type
== OBJ_BLOB
)
550 return grep_sha1(opt
, obj
->sha1
, name
, 0);
551 if (obj
->type
== OBJ_COMMIT
|| obj
->type
== OBJ_TREE
) {
552 struct tree_desc tree
;
556 data
= read_object_with_reference(obj
->sha1
, tree_type
,
559 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
560 init_tree_desc(&tree
, data
, size
);
561 hit
= grep_tree(opt
, paths
, &tree
, name
, "");
565 die("unable to grep from object of type %s", typename(obj
->type
));
568 static int context_callback(const struct option
*opt
, const char *arg
,
571 struct grep_opt
*grep_opt
= opt
->value
;
576 grep_opt
->pre_context
= grep_opt
->post_context
= 0;
579 value
= strtol(arg
, (char **)&endp
, 10);
581 return error("switch `%c' expects a numerical value",
584 grep_opt
->pre_context
= grep_opt
->post_context
= value
;
588 static int file_callback(const struct option
*opt
, const char *arg
, int unset
)
590 struct grep_opt
*grep_opt
= opt
->value
;
595 patterns
= fopen(arg
, "r");
597 die_errno("cannot open '%s'", arg
);
598 while (strbuf_getline(&sb
, patterns
, '\n') == 0) {
599 /* ignore empty line like grep does */
602 append_grep_pattern(grep_opt
, strbuf_detach(&sb
, NULL
), arg
,
603 ++lno
, GREP_PATTERN
);
610 static int not_callback(const struct option
*opt
, const char *arg
, int unset
)
612 struct grep_opt
*grep_opt
= opt
->value
;
613 append_grep_pattern(grep_opt
, "--not", "command line", 0, GREP_NOT
);
617 static int and_callback(const struct option
*opt
, const char *arg
, int unset
)
619 struct grep_opt
*grep_opt
= opt
->value
;
620 append_grep_pattern(grep_opt
, "--and", "command line", 0, GREP_AND
);
624 static int open_callback(const struct option
*opt
, const char *arg
, int unset
)
626 struct grep_opt
*grep_opt
= opt
->value
;
627 append_grep_pattern(grep_opt
, "(", "command line", 0, GREP_OPEN_PAREN
);
631 static int close_callback(const struct option
*opt
, const char *arg
, int unset
)
633 struct grep_opt
*grep_opt
= opt
->value
;
634 append_grep_pattern(grep_opt
, ")", "command line", 0, GREP_CLOSE_PAREN
);
638 static int pattern_callback(const struct option
*opt
, const char *arg
,
641 struct grep_opt
*grep_opt
= opt
->value
;
642 append_grep_pattern(grep_opt
, arg
, "-e option", 0, GREP_PATTERN
);
646 static int help_callback(const struct option
*opt
, const char *arg
, int unset
)
651 int cmd_grep(int argc
, const char **argv
, const char *prefix
)
655 int external_grep_allowed
= 1;
656 int seen_dashdash
= 0;
658 struct object_array list
= { 0, 0, NULL
};
659 const char **paths
= NULL
;
662 struct option options
[] = {
663 OPT_BOOLEAN(0, "cached", &cached
,
664 "search in index instead of in the work tree"),
666 OPT_BOOLEAN('v', "invert-match", &opt
.invert
,
667 "show non-matching lines"),
668 OPT_BIT('i', "ignore-case", &opt
.regflags
,
669 "case insensitive matching", REG_ICASE
),
670 OPT_BOOLEAN('w', "word-regexp", &opt
.word_regexp
,
671 "match patterns only at word boundaries"),
672 OPT_SET_INT('a', "text", &opt
.binary
,
673 "process binary files as text", GREP_BINARY_TEXT
),
674 OPT_SET_INT('I', NULL
, &opt
.binary
,
675 "don't match patterns in binary files",
676 GREP_BINARY_NOMATCH
),
678 OPT_BIT('E', "extended-regexp", &opt
.regflags
,
679 "use extended POSIX regular expressions", REG_EXTENDED
),
680 OPT_NEGBIT('G', "basic-regexp", &opt
.regflags
,
681 "use basic POSIX regular expressions (default)",
683 OPT_BOOLEAN('F', "fixed-strings", &opt
.fixed
,
684 "interpret patterns as fixed strings"),
686 OPT_BOOLEAN('n', NULL
, &opt
.linenum
, "show line numbers"),
687 OPT_NEGBIT('h', NULL
, &opt
.pathname
, "don't show filenames", 1),
688 OPT_BIT('H', NULL
, &opt
.pathname
, "show filenames", 1),
689 OPT_NEGBIT(0, "full-name", &opt
.relative
,
690 "show filenames relative to top directory", 1),
691 OPT_BOOLEAN('l', "files-with-matches", &opt
.name_only
,
692 "show only filenames instead of matching lines"),
693 OPT_BOOLEAN(0, "name-only", &opt
.name_only
,
694 "synonym for --files-with-matches"),
695 OPT_BOOLEAN('L', "files-without-match",
696 &opt
.unmatch_name_only
,
697 "show only the names of files without match"),
698 OPT_BOOLEAN('z', "null", &opt
.null_following_name
,
699 "print NUL after filenames"),
700 OPT_BOOLEAN('c', "count", &opt
.count
,
701 "show the number of matches instead of matching lines"),
702 OPT_SET_INT(0, "color", &opt
.color
, "highlight matches", 1),
704 OPT_CALLBACK('C', NULL
, &opt
, "n",
705 "show <n> context lines before and after matches",
707 OPT_INTEGER('B', NULL
, &opt
.pre_context
,
708 "show <n> context lines before matches"),
709 OPT_INTEGER('A', NULL
, &opt
.post_context
,
710 "show <n> context lines after matches"),
711 OPT_NUMBER_CALLBACK(&opt
, "shortcut for -C NUM",
714 OPT_CALLBACK('f', NULL
, &opt
, "file",
715 "read patterns from file", file_callback
),
716 { OPTION_CALLBACK
, 'e', NULL
, &opt
, "pattern",
717 "match <pattern>", PARSE_OPT_NONEG
, pattern_callback
},
718 { OPTION_CALLBACK
, 0, "and", &opt
, NULL
,
719 "combine patterns specified with -e",
720 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
, and_callback
},
721 OPT_BOOLEAN(0, "or", &dummy
, ""),
722 { OPTION_CALLBACK
, 0, "not", &opt
, NULL
, "",
723 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
, not_callback
},
724 { OPTION_CALLBACK
, '(', NULL
, &opt
, NULL
, "",
725 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
| PARSE_OPT_NODASH
,
727 { OPTION_CALLBACK
, ')', NULL
, &opt
, NULL
, "",
728 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
| PARSE_OPT_NODASH
,
730 OPT_BOOLEAN(0, "all-match", &opt
.all_match
,
731 "show only matches from files that match all patterns"),
734 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed
,
735 "allow calling of grep(1) (ignored by this build)"),
737 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed
,
738 "allow calling of grep(1) (default)"),
740 { OPTION_CALLBACK
, 0, "help-all", &options
, NULL
, "show usage",
741 PARSE_OPT_HIDDEN
| PARSE_OPT_NOARG
, help_callback
},
745 memset(&opt
, 0, sizeof(opt
));
746 opt
.prefix_length
= (prefix
&& *prefix
) ? strlen(prefix
) : 0;
749 opt
.pattern_tail
= &opt
.pattern_list
;
750 opt
.regflags
= REG_NEWLINE
;
752 strcpy(opt
.color_match
, GIT_COLOR_RED GIT_COLOR_BOLD
);
754 git_config(grep_config
, &opt
);
756 opt
.color
= git_use_color_default
;
759 * If there is no -- then the paths must exist in the working
760 * tree. If there is no explicit pattern specified with -e or
761 * -f, we take the first unrecognized non option to be the
762 * pattern, but then what follows it must be zero or more
763 * valid refs up to the -- (if exists), and then existing
764 * paths. If there is an explicit pattern, then the first
765 * unrecognized non option is the beginning of the refs list
766 * that continues up to the -- (if exists), and then paths.
768 argc
= parse_options(argc
, argv
, prefix
, options
, grep_usage
,
769 PARSE_OPT_KEEP_DASHDASH
|
770 PARSE_OPT_STOP_AT_NON_OPTION
|
771 PARSE_OPT_NO_INTERNAL_HELP
);
773 /* First unrecognized non-option token */
774 if (argc
> 0 && !opt
.pattern_list
) {
775 append_grep_pattern(&opt
, argv
[0], "command line", 0,
781 if (opt
.color
&& !opt
.color_external
)
782 external_grep_allowed
= 0;
783 if (!opt
.pattern_list
)
784 die("no pattern given.");
785 if ((opt
.regflags
!= REG_NEWLINE
) && opt
.fixed
)
786 die("cannot mix --fixed-strings and regexp");
787 compile_grep_patterns(&opt
);
789 /* Check revs and then paths */
790 for (i
= 0; i
< argc
; i
++) {
791 const char *arg
= argv
[i
];
792 unsigned char sha1
[20];
794 if (!get_sha1(arg
, sha1
)) {
795 struct object
*object
= parse_object(sha1
);
797 die("bad object %s", arg
);
798 add_object_array(object
, arg
, &list
);
801 if (!strcmp(arg
, "--")) {
808 /* The rest are paths */
809 if (!seen_dashdash
) {
811 for (j
= i
; j
< argc
; j
++)
812 verify_filename(prefix
, argv
[j
]);
816 paths
= get_pathspec(prefix
, argv
+ i
);
817 if (opt
.prefix_length
&& opt
.relative
) {
818 /* Make sure we do not get outside of paths */
819 for (i
= 0; paths
[i
]; i
++)
820 if (strncmp(prefix
, paths
[i
], opt
.prefix_length
))
821 die("git grep: cannot generate relative filenames containing '..'");
825 paths
= xcalloc(2, sizeof(const char *));
833 return !grep_cache(&opt
, paths
, cached
, external_grep_allowed
);
837 die("both --cached and trees are given.");
839 for (i
= 0; i
< list
.nr
; i
++) {
840 struct object
*real_obj
;
841 real_obj
= deref_tag(list
.objects
[i
].item
, NULL
, 0);
842 if (grep_object(&opt
, paths
, real_obj
, list
.objects
[i
].name
))
845 free_grep_patterns(&opt
);